botlistprojects/botbert/lib/python/business/scan_feed_bom.py

   1 """
   2  Berlin Brown
   3  6/20/2007
   4
   5  business object model
   6 """
   7 __author__ = "Berlin Brown"
   8 __version__ = "0.1"
   9
  10 import MySQLdb
  11 import cgi
  12 import random
  13
  14 import util.keyword_processor as textutils
  15 from util.text_utils import formatStrAscii, formatDescrWithSoup
  16 from botlist_business_object import BotlistBusinessObject
  17
  18 SQL_PREFIX = "'"
  19 SQL_SUFFIX = "', "
  20 SQL_EMPTY = "'', "
  21
  22 MAX_BOT_RATING = 0
  23
  24 """
  25 ------------------------------------------------
  26  Bean Definitions for Object Mapping
  27 ------------------------------------------------
  28 """
  29 class ScanFeed:
  30         def __init__(self):
  31                 self.mainUrl = None
  32                 self.urlTitle = None
  33                 self.urlDescription = None
  34                 self.urlSource = None
  35
  36 class FeedItem:
  37         def __init__(self):
  38                 self.mainUrl = None
  39                 self.urlTitle = None
  40                 self.urlDescription = None
  41                 self.urlSource = None
  42                 self.channelURL = None
  43
  44         def verify(self):
  45                 """ Because we are inserting these values into the database,
  46                 ensure that a zero length value is returned as opposed to null"""
  47                 self.channelURL =  self.channelURL and self.channelURL or ''
  48                 self.channelURL = self.channelURL.strip()
  49
  50 class EntityLink:
  51         def __init__(self):
  52                 self.mainUrl = None
  53                 self.urlTitle = None
  54                 self.urlDescription = None
  55
  56 class CountFeedItems:
  57         def __init__(self):
  58                 self.count = None
  59
  60 #--------------------------------------
  61 class ScanFeedHandler(BotlistBusinessObject):
  62         def createItem(self, url, title, descr, source, hostname):
  63                 try:
  64                         sql_set = []
  65                         sql_set.append("insert into system_feed_items(main_url, url_title, url_description, url_source, hostname) values(")
  66                         sql_set.append(SQL_PREFIX + "%s" + SQL_SUFFIX)      # main_url
  67                         sql_set.append(SQL_PREFIX + "%s" + SQL_SUFFIX)      # url_title
  68                         sql_set.append(SQL_PREFIX + "%s" + SQL_SUFFIX)      # url_descr
  69                         sql_set.append(SQL_PREFIX + "%s" + SQL_SUFFIX)      # url_source
  70                         sql_set.append(SQL_PREFIX + "%s" + SQL_PREFIX)      # hostname
  71                         sql_set.append(")")
  72                         sql_str = ''.join(sql_set)
  73                         cursor = self.conn.cursor()
  74
  75                         url_val = url.encode("utf-8").replace("'", "")
  76                         title_val =  formatStrAscii(title).replace("'", "")
  77                         descr_val =  formatDescrWithSoup(formatStrAscii(descr)).replace("'", "")
  78                         source_val = source.encode("utf-8").replace("'", "")
  79                         host_val = hostname.encode("utf-8").replace("'", "")
  80
  81                         cursor.execute(sql_str % (url_val, title_val, descr_val, source_val, host_val))
  82                         cursor.close()
  83                 except Exception, ex:
  84                         # Typically a unique id exception
  85                         raise ex
  86
  87         def updateInsertEntityProcessCount(self, link, title, user_name, hostname):
  88                 """ Complete transfer by inserting into the entity link table
  89                 for display to the user"""
  90                 data = None
  91                 try:
  92                         cursor = self.conn.cursor()
  93                         sql_set = []
  94                         sql_set.append("update system_feed_items set process_count = 1 where main_url = '%s'")
  95                         sql_str = ''.join(sql_set)
  96                         cursor.execute(sql_str % link)
  97                         sql_set = None
  98
  99                         # Filter title and keywords
 100                         cleanTitle = textutils.filterNonAscii(title.encode('utf-8', 'ignore'))
 101                         keywordTitle = textutils.createKeywords(cleanTitle)
 102
 103                         if MAX_BOT_RATING > 0:
 104                                 rnd_val = random.randint(1, MAX_BOT_RATING)
 105                         else:
 106                                 rnd_val = 0
 107
 108                         sql_set2 = []
 109                         sql_set2.append("insert into entity_links(main_url, url_title, full_name, keywords, rating, hostname, created_on) values(")
 110                         sql_set2.append(SQL_PREFIX + "%s" + SQL_SUFFIX)     # main_url
 111                         sql_set2.append(SQL_PREFIX + "%s" + SQL_SUFFIX)     # title
 112                         sql_set2.append(SQL_PREFIX + "%s" + SQL_SUFFIX)     # username
 113                         sql_set2.append(SQL_PREFIX + "%s" + SQL_SUFFIX)     # keywords
 114                         sql_set2.append("%s,")                              # rating (currently set to random value)
 115                         if hostname:
 116                                 sql_set2.append(SQL_PREFIX + "%s" + SQL_SUFFIX) # hostname
 117                         else:
 118                                 # Create a record with a NULL entry
 119                                 hostname = "NULL"
 120                                 sql_set2.append("%s,")                                                  # hostname
 121
 122                         sql_set2.append("NOW())")                           # created_on
 123                         sql_str = ''.join(sql_set2)
 124                         cursor.execute(sql_str % (link, cleanTitle, user_name, keywordTitle, rnd_val, hostname))
 125                         cursor.close()
 126                 except MySQLdb.OperationalError, message:
 127                         errorMessage = "ERR %d:\n%s" % (message[0], message[1])
 128                         return 0
 129
 130         def hasEntityLink(self, link):
 131                 data = None
 132                 try:
 133                         cursor = self.conn.cursor()
 134                         sql_set = []
 135                         sql_set.append("select count(1) from entity_links where main_url = '%s'")
 136                         sql_str = ''.join(sql_set)
 137                         cursor.execute(sql_str % (link))
 138                         data = cursor.fetchall()
 139                         cursor.close()
 140                 except MySQLdb.OperationalError, message:
 141                         errorMessage = "ERR %d:\n%s" % (message[0], message[1])
 142                         print errorMessage
 143                         return 0
 144
 145                 if data:
 146                         if len(data) > 0:
 147                                 return data[0][0]
 148                         else:
 149                                 return 0
 150                 else:
 151                         return 0
 152
 153         def listSystemFeedItems(self):
 154                 """System feed item articles have been inserted, now tranfer
 155                 these items into the entity links table"""
 156                 data = None
 157                 try:
 158                         cursor = self.conn.cursor()
 159                         sql_set = []
 160                         sql_set.append("select main_url, url_title, url_description, url_source, hostname from system_feed_items where process_count = 0 order by RAND()")
 161                         sql_str = ''.join(sql_set)
 162                         cursor.execute(sql_str)
 163                         data = cursor.fetchall()
 164                         self.fields = cursor.description
 165                         cursor.close()
 166                 except MySQLdb.OperationalError, message:
 167                         errorMessage = "ERR %d:\n%s" % (message[0], message[1])
 168                         print errorMessage
 169                         return None
 170
 171                 if not data:
 172                         raise "ERR: No data returned during listSystemFeedItems() call (try to run a scan on feeds)."
 173
 174                 transform_data = []
 175                 for row in data:
 176                         # FeedItem(rs,rs,rs,rs)
 177                         feed_item = FeedItem()
 178                         feed_item.mainUrl = cgi.escape(row[0])
 179                         feed_item.urlTitle = cgi.escape(row[1])
 180                         feed_item.urlDescription = cgi.escape(row[2])
 181                         feed_item.urlSource = cgi.escape(row[3])
 182                         if row[4]:
 183                                 feed_item.channelURL = cgi.escape(row[4])
 184
 185                         transform_data.append(feed_item)
 186
 187                 return transform_data
 188
 189         def listScanFeeds(self):
 190                 """Return a list of ScanFeed items"""
 191                 data = None
 192                 try:
 193                         cursor = self.conn.cursor()
 194                         sql_set = []
 195                         sql_set.append("select main_url, url_title, url_description, url_source from system_scan_feeds order by RAND()")
 196                         sql_str = ''.join(sql_set)
 197                         cursor.execute(sql_str)
 198                         data = cursor.fetchall()
 199                         self.fields = cursor.description
 200                         cursor.close()
 201                 except MySQLdb.OperationalError, message:
 202                         errorMessage = "ERR %d:\n%s" % (message[0], message[1])
 203                         print errorMessage
 204                         return None
 205
 206                 if not data:
 207                         return None
 208
 209                 # Transform into a list of simple objects
 210                 transform_data = []
 211                 for row in data:
 212                         feed = ScanFeed()
 213                         feed.mainUrl = cgi.escape(row[0])
 214                         feed.urlTitle = cgi.escape(row[1])
 215                         feed.urlDescription = cgi.escape(row[2])
 216                         feed.urlSource = cgi.escape(row[3])
 217                         transform_data.append(feed)
 218
 219                 return transform_data
 220
 221  # End of the file